Merge "Exit maintenance script with non-zero status if execute() returns false"
[lhc/web/wiklou.git] / maintenance / namespaceDupes.php
1 <?php
2 /**
3 * Check for articles to fix after adding/deleting namespaces
4 *
5 * Copyright © 2005-2007 Brion Vibber <brion@pobox.com>
6 * https://www.mediawiki.org/
7 *
8 * This program is free software; you can redistribute it and/or modify
9 * it under the terms of the GNU General Public License as published by
10 * the Free Software Foundation; either version 2 of the License, or
11 * (at your option) any later version.
12 *
13 * This program is distributed in the hope that it will be useful,
14 * but WITHOUT ANY WARRANTY; without even the implied warranty of
15 * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the
16 * GNU General Public License for more details.
17 *
18 * You should have received a copy of the GNU General Public License along
19 * with this program; if not, write to the Free Software Foundation, Inc.,
20 * 51 Franklin Street, Fifth Floor, Boston, MA 02110-1301, USA.
21 * http://www.gnu.org/copyleft/gpl.html
22 *
23 * @file
24 * @ingroup Maintenance
25 */
26
27 require_once __DIR__ . '/Maintenance.php';
28
29 use MediaWiki\Linker\LinkTarget;
30 use MediaWiki\MediaWikiServices;
31 use Wikimedia\Rdbms\ResultWrapper;
32 use Wikimedia\Rdbms\IMaintainableDatabase;
33
34 /**
35 * Maintenance script that checks for articles to fix after
36 * adding/deleting namespaces.
37 *
38 * @ingroup Maintenance
39 */
40 class NamespaceDupes extends Maintenance {
41
42 /**
43 * @var IMaintainableDatabase
44 */
45 protected $db;
46
47 private $resolvablePages = 0;
48 private $totalPages = 0;
49
50 private $resolvableLinks = 0;
51 private $totalLinks = 0;
52
53 public function __construct() {
54 parent::__construct();
55 $this->addDescription( 'Find and fix pages affected by namespace addition/removal' );
56 $this->addOption( 'fix', 'Attempt to automatically fix errors' );
57 $this->addOption( 'merge', "Instead of renaming conflicts, do a history merge with " .
58 "the correct title" );
59 $this->addOption( 'add-suffix', "Dupes will be renamed with correct namespace with " .
60 "<text> appended after the article name", false, true );
61 $this->addOption( 'add-prefix', "Dupes will be renamed with correct namespace with " .
62 "<text> prepended before the article name", false, true );
63 $this->addOption( 'source-pseudo-namespace', "Move all pages with the given source " .
64 "prefix (with an implied colon following it). If --dest-namespace is not specified, " .
65 "the colon will be replaced with a hyphen.",
66 false, true );
67 $this->addOption( 'dest-namespace', "In combination with --source-pseudo-namespace, " .
68 "specify the namespace ID of the destination.", false, true );
69 $this->addOption( 'move-talk', "If this is specified, pages in the Talk namespace that " .
70 "begin with a conflicting prefix will be renamed, for example " .
71 "Talk:File:Foo -> File_Talk:Foo" );
72 }
73
74 public function execute() {
75 $this->db = $this->getDB( DB_MASTER );
76
77 $options = [
78 'fix' => $this->hasOption( 'fix' ),
79 'merge' => $this->hasOption( 'merge' ),
80 'add-suffix' => $this->getOption( 'add-suffix', '' ),
81 'add-prefix' => $this->getOption( 'add-prefix', '' ),
82 'move-talk' => $this->hasOption( 'move-talk' ),
83 'source-pseudo-namespace' => $this->getOption( 'source-pseudo-namespace', '' ),
84 'dest-namespace' => intval( $this->getOption( 'dest-namespace', 0 ) ) ];
85
86 if ( $options['source-pseudo-namespace'] !== '' ) {
87 $retval = $this->checkPrefix( $options );
88 } else {
89 $retval = $this->checkAll( $options );
90 }
91
92 if ( $retval ) {
93 $this->output( "\nLooks good!\n" );
94 } else {
95 $this->output( "\nOh noeees\n" );
96 }
97 }
98
99 /**
100 * Check all namespaces
101 *
102 * @param array $options Associative array of validated command-line options
103 *
104 * @return bool
105 */
106 private function checkAll( $options ) {
107 global $wgContLang, $wgNamespaceAliases, $wgCapitalLinks;
108
109 $spaces = [];
110
111 // List interwikis first, so they'll be overridden
112 // by any conflicting local namespaces.
113 foreach ( $this->getInterwikiList() as $prefix ) {
114 $name = $wgContLang->ucfirst( $prefix );
115 $spaces[$name] = 0;
116 }
117
118 // Now pull in all canonical and alias namespaces...
119 foreach ( MWNamespace::getCanonicalNamespaces() as $ns => $name ) {
120 // This includes $wgExtraNamespaces
121 if ( $name !== '' ) {
122 $spaces[$name] = $ns;
123 }
124 }
125 foreach ( $wgContLang->getNamespaces() as $ns => $name ) {
126 if ( $name !== '' ) {
127 $spaces[$name] = $ns;
128 }
129 }
130 foreach ( $wgNamespaceAliases as $name => $ns ) {
131 $spaces[$name] = $ns;
132 }
133 foreach ( $wgContLang->getNamespaceAliases() as $name => $ns ) {
134 $spaces[$name] = $ns;
135 }
136
137 // We'll need to check for lowercase keys as well,
138 // since we're doing case-sensitive searches in the db.
139 foreach ( $spaces as $name => $ns ) {
140 $moreNames = [];
141 $moreNames[] = $wgContLang->uc( $name );
142 $moreNames[] = $wgContLang->ucfirst( $wgContLang->lc( $name ) );
143 $moreNames[] = $wgContLang->ucwords( $name );
144 $moreNames[] = $wgContLang->ucwords( $wgContLang->lc( $name ) );
145 $moreNames[] = $wgContLang->ucwordbreaks( $name );
146 $moreNames[] = $wgContLang->ucwordbreaks( $wgContLang->lc( $name ) );
147 if ( !$wgCapitalLinks ) {
148 foreach ( $moreNames as $altName ) {
149 $moreNames[] = $wgContLang->lcfirst( $altName );
150 }
151 $moreNames[] = $wgContLang->lcfirst( $name );
152 }
153 foreach ( array_unique( $moreNames ) as $altName ) {
154 if ( $altName !== $name ) {
155 $spaces[$altName] = $ns;
156 }
157 }
158 }
159
160 // Sort by namespace index, and if there are two with the same index,
161 // break the tie by sorting by name
162 $origSpaces = $spaces;
163 uksort( $spaces, function ( $a, $b ) use ( $origSpaces ) {
164 return $origSpaces[$a] <=> $origSpaces[$b]
165 ?: $a <=> $b;
166 } );
167
168 $ok = true;
169 foreach ( $spaces as $name => $ns ) {
170 $ok = $this->checkNamespace( $ns, $name, $options ) && $ok;
171 }
172
173 $this->output( "{$this->totalPages} pages to fix, " .
174 "{$this->resolvablePages} were resolvable.\n\n" );
175
176 foreach ( $spaces as $name => $ns ) {
177 if ( $ns != 0 ) {
178 /* Fix up link destinations for non-interwiki links only.
179 *
180 * For example if a page has [[Foo:Bar]] and then a Foo namespace
181 * is introduced, pagelinks needs to be updated to have
182 * page_namespace = NS_FOO.
183 *
184 * If instead an interwiki prefix was introduced called "Foo",
185 * the link should instead be moved to the iwlinks table. If a new
186 * language is introduced called "Foo", or if there is a pagelink
187 * [[fr:Bar]] when interlanguage magic links are turned on, the
188 * link would have to be moved to the langlinks table. Let's put
189 * those cases in the too-hard basket for now. The consequences are
190 * not especially severe.
191 * @fixme Handle interwiki links, and pagelinks to Category:, File:
192 * which probably need reparsing.
193 */
194
195 $this->checkLinkTable( 'pagelinks', 'pl', $ns, $name, $options );
196 $this->checkLinkTable( 'templatelinks', 'tl', $ns, $name, $options );
197
198 // The redirect table has interwiki links randomly mixed in, we
199 // need to filter those out. For example [[w:Foo:Bar]] would
200 // have rd_interwiki=w and rd_namespace=0, which would match the
201 // query for a conflicting namespace "Foo" if filtering wasn't done.
202 $this->checkLinkTable( 'redirect', 'rd', $ns, $name, $options,
203 [ 'rd_interwiki' => null ] );
204 $this->checkLinkTable( 'redirect', 'rd', $ns, $name, $options,
205 [ 'rd_interwiki' => '' ] );
206 }
207 }
208
209 $this->output( "{$this->totalLinks} links to fix, " .
210 "{$this->resolvableLinks} were resolvable.\n" );
211
212 return $ok;
213 }
214
215 /**
216 * Get the interwiki list
217 *
218 * @return array
219 */
220 private function getInterwikiList() {
221 $result = MediaWikiServices::getInstance()->getInterwikiLookup()->getAllPrefixes();
222 $prefixes = [];
223 foreach ( $result as $row ) {
224 $prefixes[] = $row['iw_prefix'];
225 }
226
227 return $prefixes;
228 }
229
230 /**
231 * Check a given prefix and try to move it into the given destination namespace
232 *
233 * @param int $ns Destination namespace id
234 * @param string $name
235 * @param array $options Associative array of validated command-line options
236 * @return bool
237 */
238 private function checkNamespace( $ns, $name, $options ) {
239 $targets = $this->getTargetList( $ns, $name, $options );
240 $count = $targets->numRows();
241 $this->totalPages += $count;
242 if ( $count == 0 ) {
243 return true;
244 }
245
246 $dryRunNote = $options['fix'] ? '' : ' DRY RUN ONLY';
247
248 $ok = true;
249 foreach ( $targets as $row ) {
250 // Find the new title and determine the action to take
251
252 $newTitle = $this->getDestinationTitle( $ns, $name,
253 $row->page_namespace, $row->page_title, $options );
254 $logStatus = false;
255 if ( !$newTitle ) {
256 $logStatus = 'invalid title';
257 $action = 'abort';
258 } elseif ( $newTitle->exists() ) {
259 if ( $options['merge'] ) {
260 if ( $this->canMerge( $row->page_id, $newTitle, $logStatus ) ) {
261 $action = 'merge';
262 } else {
263 $action = 'abort';
264 }
265 } elseif ( $options['add-prefix'] == '' && $options['add-suffix'] == '' ) {
266 $action = 'abort';
267 $logStatus = 'dest title exists and --add-prefix not specified';
268 } else {
269 $newTitle = $this->getAlternateTitle( $newTitle, $options );
270 if ( !$newTitle ) {
271 $action = 'abort';
272 $logStatus = 'alternate title is invalid';
273 } elseif ( $newTitle->exists() ) {
274 $action = 'abort';
275 $logStatus = 'title conflict';
276 } else {
277 $action = 'move';
278 $logStatus = 'alternate';
279 }
280 }
281 } else {
282 $action = 'move';
283 $logStatus = 'no conflict';
284 }
285
286 // Take the action or log a dry run message
287
288 $logTitle = "id={$row->page_id} ns={$row->page_namespace} dbk={$row->page_title}";
289 $pageOK = true;
290
291 switch ( $action ) {
292 case 'abort':
293 $this->output( "$logTitle *** $logStatus\n" );
294 $pageOK = false;
295 break;
296 case 'move':
297 $this->output( "$logTitle -> " .
298 $newTitle->getPrefixedDBkey() . " ($logStatus)$dryRunNote\n" );
299
300 if ( $options['fix'] ) {
301 $pageOK = $this->movePage( $row->page_id, $newTitle );
302 }
303 break;
304 case 'merge':
305 $this->output( "$logTitle => " .
306 $newTitle->getPrefixedDBkey() . " (merge)$dryRunNote\n" );
307
308 if ( $options['fix'] ) {
309 $pageOK = $this->mergePage( $row, $newTitle );
310 }
311 break;
312 }
313
314 if ( $pageOK ) {
315 $this->resolvablePages++;
316 } else {
317 $ok = false;
318 }
319 }
320
321 return $ok;
322 }
323
324 /**
325 * Check and repair the destination fields in a link table
326 * @param string $table The link table name
327 * @param string $fieldPrefix The field prefix in the link table
328 * @param int $ns Destination namespace id
329 * @param string $name
330 * @param array $options Associative array of validated command-line options
331 * @param array $extraConds Extra conditions for the SQL query
332 */
333 private function checkLinkTable( $table, $fieldPrefix, $ns, $name, $options,
334 $extraConds = []
335 ) {
336 $batchConds = [];
337 $fromField = "{$fieldPrefix}_from";
338 $namespaceField = "{$fieldPrefix}_namespace";
339 $titleField = "{$fieldPrefix}_title";
340 $batchSize = 500;
341 while ( true ) {
342 $res = $this->db->select(
343 $table,
344 [ $fromField, $namespaceField, $titleField ],
345 array_merge( $batchConds, $extraConds, [
346 $namespaceField => 0,
347 $titleField . $this->db->buildLike( "$name:", $this->db->anyString() )
348 ] ),
349 __METHOD__,
350 [
351 'ORDER BY' => [ $titleField, $fromField ],
352 'LIMIT' => $batchSize
353 ]
354 );
355
356 if ( $res->numRows() == 0 ) {
357 break;
358 }
359 foreach ( $res as $row ) {
360 $logTitle = "from={$row->$fromField} ns={$row->$namespaceField} " .
361 "dbk={$row->$titleField}";
362 $destTitle = $this->getDestinationTitle( $ns, $name,
363 $row->$namespaceField, $row->$titleField, $options );
364 $this->totalLinks++;
365 if ( !$destTitle ) {
366 $this->output( "$table $logTitle *** INVALID\n" );
367 continue;
368 }
369 $this->resolvableLinks++;
370 if ( !$options['fix'] ) {
371 $this->output( "$table $logTitle -> " .
372 $destTitle->getPrefixedDBkey() . " DRY RUN\n" );
373 continue;
374 }
375
376 $this->db->update( $table,
377 // SET
378 [
379 $namespaceField => $destTitle->getNamespace(),
380 $titleField => $destTitle->getDBkey()
381 ],
382 // WHERE
383 [
384 $namespaceField => 0,
385 $titleField => $row->$titleField,
386 $fromField => $row->$fromField
387 ],
388 __METHOD__,
389 [ 'IGNORE' ]
390 );
391 $this->output( "$table $logTitle -> " .
392 $destTitle->getPrefixedDBkey() . "\n" );
393 }
394 $encLastTitle = $this->db->addQuotes( $row->$titleField );
395 $encLastFrom = $this->db->addQuotes( $row->$fromField );
396
397 $batchConds = [
398 "$titleField > $encLastTitle " .
399 "OR ($titleField = $encLastTitle AND $fromField > $encLastFrom)" ];
400
401 wfWaitForSlaves();
402 }
403 }
404
405 /**
406 * Move the given pseudo-namespace, either replacing the colon with a hyphen
407 * (useful for pseudo-namespaces that conflict with interwiki links) or move
408 * them to another namespace if specified.
409 * @param array $options Associative array of validated command-line options
410 * @return bool
411 */
412 private function checkPrefix( $options ) {
413 $prefix = $options['source-pseudo-namespace'];
414 $ns = $options['dest-namespace'];
415 $this->output( "Checking prefix \"$prefix\" vs namespace $ns\n" );
416
417 return $this->checkNamespace( $ns, $prefix, $options );
418 }
419
420 /**
421 * Find pages in main and talk namespaces that have a prefix of the new
422 * namespace so we know titles that will need migrating
423 *
424 * @param int $ns Destination namespace id
425 * @param string $name Prefix that is being made a namespace
426 * @param array $options Associative array of validated command-line options
427 *
428 * @return ResultWrapper
429 */
430 private function getTargetList( $ns, $name, $options ) {
431 if ( $options['move-talk'] && MWNamespace::isSubject( $ns ) ) {
432 $checkNamespaces = [ NS_MAIN, NS_TALK ];
433 } else {
434 $checkNamespaces = NS_MAIN;
435 }
436
437 return $this->db->select( 'page',
438 [
439 'page_id',
440 'page_title',
441 'page_namespace',
442 ],
443 [
444 'page_namespace' => $checkNamespaces,
445 'page_title' . $this->db->buildLike( "$name:", $this->db->anyString() ),
446 ],
447 __METHOD__
448 );
449 }
450
451 /**
452 * Get the preferred destination title for a given target page.
453 * @param int $ns The destination namespace ID
454 * @param string $name The conflicting prefix
455 * @param int $sourceNs The source namespace
456 * @param int $sourceDbk The source DB key (i.e. page_title)
457 * @param array $options Associative array of validated command-line options
458 * @return Title|false
459 */
460 private function getDestinationTitle( $ns, $name, $sourceNs, $sourceDbk, $options ) {
461 $dbk = substr( $sourceDbk, strlen( "$name:" ) );
462 if ( $ns == 0 ) {
463 // An interwiki; try an alternate encoding with '-' for ':'
464 $dbk = "$name-" . $dbk;
465 }
466 $destNS = $ns;
467 if ( $sourceNs == NS_TALK && MWNamespace::isSubject( $ns ) ) {
468 // This is an associated talk page moved with the --move-talk feature.
469 $destNS = MWNamespace::getTalk( $destNS );
470 }
471 $newTitle = Title::makeTitleSafe( $destNS, $dbk );
472 if ( !$newTitle || !$newTitle->canExist() ) {
473 return false;
474 }
475 return $newTitle;
476 }
477
478 /**
479 * Get an alternative title to move a page to. This is used if the
480 * preferred destination title already exists.
481 *
482 * @param LinkTarget $linkTarget
483 * @param array $options Associative array of validated command-line options
484 * @return Title|bool
485 */
486 private function getAlternateTitle( LinkTarget $linkTarget, $options ) {
487 $prefix = $options['add-prefix'];
488 $suffix = $options['add-suffix'];
489 if ( $prefix == '' && $suffix == '' ) {
490 return false;
491 }
492 while ( true ) {
493 $dbk = $prefix . $linkTarget->getDBkey() . $suffix;
494 $title = Title::makeTitleSafe( $linkTarget->getNamespace(), $dbk );
495 if ( !$title ) {
496 return false;
497 }
498 if ( !$title->exists() ) {
499 return $title;
500 }
501 }
502 }
503
504 /**
505 * Move a page
506 *
507 * @param integer $id The page_id
508 * @param LinkTarget $newLinkTarget The new title link target
509 * @return bool
510 */
511 private function movePage( $id, LinkTarget $newLinkTarget ) {
512 $this->db->update( 'page',
513 [
514 "page_namespace" => $newLinkTarget->getNamespace(),
515 "page_title" => $newLinkTarget->getDBkey(),
516 ],
517 [
518 "page_id" => $id,
519 ],
520 __METHOD__ );
521
522 // Update *_from_namespace in links tables
523 $fromNamespaceTables = [
524 [ 'pagelinks', 'pl' ],
525 [ 'templatelinks', 'tl' ],
526 [ 'imagelinks', 'il' ] ];
527 foreach ( $fromNamespaceTables as $tableInfo ) {
528 list( $table, $fieldPrefix ) = $tableInfo;
529 $this->db->update( $table,
530 // SET
531 [ "{$fieldPrefix}_from_namespace" => $newLinkTarget->getNamespace() ],
532 // WHERE
533 [ "{$fieldPrefix}_from" => $id ],
534 __METHOD__ );
535 }
536
537 return true;
538 }
539
540 /**
541 * Determine if we can merge a page.
542 * We check if an inaccessible revision would become the latest and
543 * deny the merge if so -- it's theoretically possible to update the
544 * latest revision, but opens a can of worms -- search engine updates,
545 * recentchanges review, etc.
546 *
547 * @param integer $id The page_id
548 * @param LinkTarget $linkTarget The new link target
549 * @param string $logStatus This is set to the log status message on failure
550 * @return bool
551 */
552 private function canMerge( $id, LinkTarget $linkTarget, &$logStatus ) {
553 $latestDest = Revision::newFromTitle( $linkTarget, 0, Revision::READ_LATEST );
554 $latestSource = Revision::newFromPageId( $id, 0, Revision::READ_LATEST );
555 if ( $latestSource->getTimestamp() > $latestDest->getTimestamp() ) {
556 $logStatus = 'cannot merge since source is later';
557 return false;
558 } else {
559 return true;
560 }
561 }
562
563 /**
564 * Merge page histories
565 *
566 * @param stdClass $row Page row
567 * @param Title $newTitle The new title
568 * @return bool
569 */
570 private function mergePage( $row, Title $newTitle ) {
571 $id = $row->page_id;
572
573 // Construct the WikiPage object we will need later, while the
574 // page_id still exists. Note that this cannot use makeTitleSafe(),
575 // we are deliberately constructing an invalid title.
576 $sourceTitle = Title::makeTitle( $row->page_namespace, $row->page_title );
577 $sourceTitle->resetArticleID( $id );
578 $wikiPage = new WikiPage( $sourceTitle );
579 $wikiPage->loadPageData( 'fromdbmaster' );
580
581 $destId = $newTitle->getArticleID();
582 $this->beginTransaction( $this->db, __METHOD__ );
583 $this->db->update( 'revision',
584 // SET
585 [ 'rev_page' => $destId ],
586 // WHERE
587 [ 'rev_page' => $id ],
588 __METHOD__ );
589
590 $this->db->delete( 'page', [ 'page_id' => $id ], __METHOD__ );
591
592 $this->commitTransaction( $this->db, __METHOD__ );
593
594 /* Call LinksDeletionUpdate to delete outgoing links from the old title,
595 * and update category counts.
596 *
597 * Calling external code with a fake broken Title is a fairly dubious
598 * idea. It's necessary because it's quite a lot of code to duplicate,
599 * but that also makes it fragile since it would be easy for someone to
600 * accidentally introduce an assumption of title validity to the code we
601 * are calling.
602 */
603 DeferredUpdates::addUpdate( new LinksDeletionUpdate( $wikiPage ) );
604 DeferredUpdates::doUpdates();
605
606 return true;
607 }
608 }
609
610 $maintClass = NamespaceDupes::class;
611 require_once RUN_MAINTENANCE_IF_MAIN;